---
license: apache-2.0
datasets:
- Quest-AI/quest-corruption-truncated4grpo-2k-dataset-v1
base_model:
- Quest-AI/quest-corruption-14b-s110-r3
---

Alternative version of my prior 14b GRPO attempt with a higher learning rate, on a limited selection of data (2k entries instead of 6k).

![image/png](https://cdn-uploads.huggingface.co/production/uploads/6491e00e057b0928b3e07b75/pyrIVlW6jSVRxq803CjWk.png)

Wandb with hparams and code in the files section can be found here:

[https://wandb.ai/kalomaze/verifiers-examples/runs/mayn2ctv?nw=nwuserkalomaze](https://wandb.ai/kalomaze/verifiers-examples/runs/mayn2ctv?nw=nwuserkalomaze)